You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
If ProofCollection proving can run with just 8Gb ram then we should lower the min requirement (currently 16Gb) to 8Gb in CliArgs::estimate_proving_requirements(). That enables more people to run nodes that can initiate tx. Particularly on lower end laptops and VPS.
TVM_LDE_TRACE="no_cache" should be on by default for ProofCollection jobs. (And off by default for SingleProof jobs.)
It is an env variable that can be set before invoking the triton-vm-prove process. If it is not set, the (faster) memory-greedy code path is taken. If not set, it is memory-frugal but slower.
enhancement: It could possibly be turned into a neptune-core cli flag as well, and pass that option to the proving job.
The text was updated successfully, but these errors were encountered:
If ProofCollection proving can run with just 8Gb ram then we should lower the min requirement (currently 16Gb) to 8Gb in CliArgs::estimate_proving_requirements(). That enables more people to run nodes that can initiate tx. Particularly on lower end laptops and VPS.
TVM_LDE_TRACE="no_cache" should be on by default for ProofCollection jobs. (And off by default for SingleProof jobs.)
It is an env variable that can be set before invoking the triton-vm-prove process. If it is not set, the (faster) memory-greedy code path is taken. If not set, it is memory-frugal but slower.
enhancement: It could possibly be turned into a neptune-core cli flag as well, and pass that option to the proving job.
The text was updated successfully, but these errors were encountered: